Urban Villages in China: A 2008 Survey of Migrant … ET AL. 427 better understanding of the housing demand a nd housing choice behaviors of migrants is cru-cial for developing effective

Zheng.fm Page 425 Monday, July 13, 2009 10:14 PM

Urban Villages in China: A 2008 Survey of Migrant Settlements in Beijing

Siqi Zheng, Fenjie Long, C. Cindy Fan, and Yizhen Gu1

Abstract: A team of Beijing-based urban planning specialists is joined by a noted Americangeographer to present the results and analyze their 2008 survey of migrant settlements inChina’s capital city. The paper examines the living and work conditions as well as housingconsumption behavior of migrants in Chinese cities, focusing on chengzhongcun or urban vil-lages—rural settlements that have been transformed into poor living spaces for migrant work-ers. It finds that although migrant workers are willing to pay the same or higher rent per unitof space, they consume much smaller dwelling spaces than local residents. Estimations of theMincerian wage equation and of a housing demand equation show that migrants’ small spaceconsumption is a function not only of low income but also of a reluctance to spend their earn-ings in the city. The findings reinforce the notion that migrant workers consider the city as aplace to work rather than a home in which to live. Journal of Economic Literature, Classifica-tion Numbers: J610, O150, R210, R230. 14 figures, 5 tables, 34 references, 1 appendix. Keywords: China, Beijing, urban villages, chengzhongcun, rural-urban migration, urbanization,housing, rent, rural settlements.

INTRODUCTION

ver the past three decades, China has experienced rapid urban growth and massiverural-urban migration. The share of urban population increased from 21 percent in

1982 to 45 in 2007 and is expected to exceed 50 percent by 2015 (Zhou and Ma, 2005;Guangming ribao, 2006; National Bureau, 2008). In 2007, there were 36 cities in China witha population of two million or more (National Bureau, 2008). Rural-urban migration isresponsible for roughly 70 percent of the country’s urban population growth (Zhang andSong, 2003).

Such rapid urbanization has transformed the spatial and social landscapes of Chinese cit-ies. One of the most prominent imprints of rural-urban migrants is chengzhongcun—literally“villages in the city,” also translated as “villages amid the city,” “villages encircled by thecity,” and “urban villages” (e.g., Zhang et al., 2003; Wu, 2007; Tian, 2008). Whichever

1Respectively, Associate Professor, Institute of Real Estate Studies, Tsinghua University, Beijing 100084,China ([email protected]); Associate Professor, Institute of Real Estate Studies, Tsinghua University,Beijing 100084, China ([email protected]); Professor of Geography, Department of Geography, University ofCalifornia, Los Angeles, CA 90095-1524 ([email protected]); and Engineer, Beijing Municipal Institute of CityPlanning and Design, No. 60, South Lishi Road, Beijing 100045, China ([email protected]). The authors wishto thank Yang Cao, Ning Wang, Cong Sun, and Chase Langford for providing excellent research assistance.Research presented in this paper was supported in part by a National Science Foundation award (BCS-0455107),funding from the UCLA Academic Senate, UCLA Asian American Studies Center, National Natural Science Foun-dation of China (70603017) and by the Beijing Municipal Institute of City Planning and Design.

O

425

Eurasian Geography and Economics, 2009, 50, No. 4, pp. 425–446. DOI: 10.2747/1539-7216.50.4.425Copyright © 2009 by Bellwether Publishing, Ltd. All rights reserved.

426 EURASIAN GEOGRAPHY AND ECONOMICS


translation is used, these terms describe rural villages that have been surrounded or otherwiseencroached upon by urban expansion. For the sake of simplicity, we use in this paper theterm “urban villages” to refer to chengzhongcun.2

While farmland in villages on city fringes may have been acquired by city governments,land for housing continues to be owned by village collectives and is allocated to village resi-dents. However, many urban village residents have given up farming and have instead builtor expanded housing to rent to migrants. To these villagers, rent has replaced agriculture asthe main source of income. Urban villages can be found most commonly on the fringes ofcities that have experienced significant expansion and received large numbers of migrants.These urban fringes surrounding Chinese cities have been referred to variously as urban out-skirts, peri-urban areas, and suburban areas (Lin, 2006; McGee et al., 2007; Zhou and Logan,2008). Hundreds of urban villages exist in large cities such as Guangzhou and Shenzhen.According to You-tien Hsing (2009), there were 139 urban villages in Guangzhou in 2006,and urban villages in Guangzhou and Shenzhen make up more than, respectively, 20 and 60percent of their planned areas providing homes to 80 percent of migrants in these cities.

The emergence of urban villages is an outcome not only of rapid urbanization but also ofthe persistent divides between rural and urban citizenships and between rural and urbanadministrations in China (Zhang and Song, 2003; Song et al., 2008). From the demand side,millions of rural migrants working in cities have generated enormous demand for inexpen-sive housing. Most rural migrants are excluded from the formal housing market because: (1)without urban hukou (household registration) they are not eligible for low-cost affordablehousing subsidized by city governments; and (2) they cannot afford “commodity housing” inthe private housing market. Therefore, migrants most commonly live in employer-providedhousing such as factory dorms or rent rooms in urban villages.

From the supply side, as cities expand, municipal governments are motivated to acquirefarmland for urban use. To minimize compensation for villagers’ housing and relocation andto ease the process of land acquisition, city governments tend to acquire farmland only andleave or return the land designated for housing—“reserved housing sites” (zhaijidi)—to vil-lage collectives (Hsing, 2009). Typically, these “rural” land parcels are spatially scatteredand receive no public services from city governments. But because villagers are not requiredto pay a land lease fee for using these parcels, the cost and rent of housing built on such landis low and especially attractive to migrant workers who cannot afford high rent. To maximizerental income, villagers build high-density houses and add floors and structures haphazardlyand even illegally, resulting in slum-like living environments (Tang and Chung, 2002).Unlike shantytowns in Latin America, however, urban villages in Chinese cities are not aresult of land invasion and self-constructed housing by migrants (Mobrand, 2006). Rather,they represent a match between migrants’ demand for cheap housing and the supply of low-cost housing in villages encroached upon by urban expansion.

Notwithstanding their positive role in providing housing to migrants, urban villages areof great concern to city governments. High crime rates, inadequate infrastructure and ser-vices, and poor living conditions are just some of the problems in urban villages that threatenpublic security and management (Zhang, 2002). Yet eliminating urban villages altogether isnot a sustainable solution, unless provisions to re-house migrants are in place. Therefore, a

2Note that our usage of the term “urban village” differs from the planning concept influenced by Jane Jacobs(1961) that advocates urban living marked by self-containment, community interaction, and reduced reliance onautomobiles. The latter conception of urban village is seen as an alternative to the urban sprawl that since WWII hascharacterized many Western cities.

ZHENG ET AL. 427


better understanding of the housing demand and housing choice behaviors of migrants is cru-cial for developing effective strategies to better house migrants.

Drawing from a survey of 50 urban villages and 756 migrants in Beijing conducted inSeptember 2008, this paper aims at providing fresh information on the living and employ-ment conditions of migrants in the city, with a special focus on understanding their rationalefor choosing urban villages as places to live. In the next section, we describe urban villagesin Beijing and our survey. This is followed by a descriptive and statistical summary of thesurveyed urban villages and migrants. We then turn to migrants’ housing condition and hous-ing consumption behavior, followed by econometric models of wage and housing demand.The concluding section summarizes our findings and discusses their policy implications.

URBAN VILLAGES IN BEIJING

As defined by the Beijing Municipal Commission of Urban Planning, the Beijing Metro-politan Area (BMA) is 1,086 km2 in size (see the area shaded in light grey in Fig. 1) and con-sists of four urban districts (Dongcheng, Xicheng, Chongwen, and Xuanwu) and parts of fivesuburban districts (Chaoyang, Haidian, Fengtai, Shijingshan, and Changping).

Because urban villages are not formally considered part of the urban economy—andthus excluded from urban statistics collection—information about their number and spatialdistribution is not publicly available. Working with Beijing Municipal Land Authority, we

Fig. 1. Distribution of urban villages in the Beijing Metropolitan Area. Source: The Second LandSurvey of Beijing, conducted by the Beijing Municipal Land Authority in 2007.



have obtained a detailed list of urban villages and the land parcels occupied by these villages.As of 2008, there were 867 urban villages in BMA, mostly located in the suburban districts(see the areas shaded black in Fig. 1). These urban villages occupy 181 km2, accounting for49.5 percent of the total residential land in BMA. Despite their informal status in the housingmarket, urban villages clearly play a prominent role in the provision of housing in Beijing.

Our empirical analysis is based on a questionnaire survey conducted in September 2008and administered by Beijing Municipal Institute of Urban Planning and Design and TsinghuaUniversity’s Institute of Real Estate Studies (hereafter Beijing Urban Village Survey). Weemployed a two-stage sampling method. In the first stage, 50 urban villages were selectedrandomly from the total of 867 in BMA (Fig. 2 and Table A1 in the Appendix). In the secondstage, we selected 15–20 migrants in each of the 50 urban villages. To increase representa-tiveness, we selected migrants from different parts of the villages so that they were approxi-mately evenly distributed spatially. Considering that most migrants have low levels ofeducational attainment, our interviewers asked them questions and filled out their question-naires rather than asking them to do it themselves. The survey yielded 756 completed andvalid questionnaires. In each urban village, we also interviewed village officials for generalinformation.

Information from village officials shows that in most of the 50 selected urban villages,migrants (individuals who do not have urban or rural hukou in Beijing) outnumber natives(individuals who have urban or rural hukou in Beijing) (Fig. 2 and Table A1). In fact, most

Fig. 2. The 50 sampled sites in the Beijing Urban Village Survey. Grey areas refer to the four sub-urban districts adjacent to Beijing’s four urban districts. Open (white) circles refer to urban villages inwhich numbers for migrant and native populations are not available (see Table A1).

ZHENG ET AL. 429


villages have a large migrant–native ratio, indicating that the majority of people living in theurban villages are renters. Among the 50 urban villages we surveyed, there are 5.2 migrantsfor every native. The migrant–native ratio varies among villages. Among the several urbanvillages in the north and east that have especially large ratios, some are in relative proximityto low-skill industrial clusters that employ large numbers of migrants and are thereforeattractive to them as places to live.

DEMOGRAPHIC AND ECONOMIC CHARACTERISTICS OF MIGRANTS

Most migrants in urban villages are young men, who have low levels of educationalattainment and low income; they work in the tertiary sector, hold rural hukou, and are fromadjacent or poor provinces. In our survey, men accounted for 66.9 percent of migrants resid-ing in the urban villages, a proportion much higher than the 50.8 percent of men amongBeijing’s permanent population as of 2007 and also higher than the 62.4 percent of menamong migrants in the BMA overall (defined as individuals who moved to Beijing within thelast five years) based on the 2000 Census (National Bureau, 2002) (Table 1). Figure 3 com-pares the age distribution of the surveyed migrants with that of Beijing’s permanent popula-tion in 2007. Migrants are highly concentrated in the 20–39 age cohort, with a mean age of34, compared to a mean age of 37.8 among Beijing’s permanent population. The dominance

Table 1. Gender Composition of Migrants and Beijing’s Population, 2000–2008 (percent)

Inhabitants of Beijing Male Female

Migrants in 2008 survey 66.9 33.1Migrants in 2000 Census 62.4 37.6Permanent residents in 2007 50.8 49.2

Sources: Compiled by authors from Beijing Urban Village Survey; National Bureau, 2002; and Beijing Bureau,2008.

Fig. 3. Age distribution of migrants and Beijing’s permanent population. Sources: Beijing UrbanVillage Survey; Beijing Bureau (2008).



of young males among migrants is consistent with the notion that their main purpose of stay-ing in cities is to work.

Figure 4 compares the educational attainment of migrants in our survey with that ofBeijing’s permanent population. Among migrants, 73.1 percent have junior high school orlower levels of educational attainment, in comparison with 47.0 percent among Beijing’spopulation. A low level of educational attainment is among the factors contributing to themigrants’ low income and their concentrated employment in low-skilled sectors.

The mean monthly income of migrants in the survey is 1,984 yuan, compared to a meanincome of 3,876 yuan for the labor force in Beijing as of 2007 (Beijing Bureau, 2008). As isevident from Figure 5, the income distribution of migrants peaks at around 2000 yuan beforefalling off at higher income levels.

Among the migrants in the survey, 72.6 percent work in the tertiary sector,3 a slightly largershare than in Beijing’s labor force as a whole (69.3 percent) (Beijing Bureau, 2008). Figure 6shows that for every major occupational category migrants earn less than the average forBeijing’s permanent labor force. Such an income discrepancy suggests that: (1) migrantsconcentrate in low-skilled jobs within each industry; and (2) they encounter discrimination inthe labor market (Fan, 2001, 2002).

The vast majority—85 percent—of the migrants in our survey hold rural hukou. The resthave urban hukou from places other than Beijing, with the majority of being from small

3The tertiary sector as defined by Chinese statistical agencies includes the following activities: transport, stor-age, and postal services; information transmission, computer services, and software; wholesale and retail trade; hoteland restaurants; financial intermediation; real estate; leasing and business services; scientific research, technical ser-vices, and geologic prospecting; management of water resources, the environment, and public facilities; householdand other services; education; health, social security, and social welfare; culture, sports, and entertainment; and pub-lic management and social organization.

Fig. 4. Educational attainment of migrants and Beijing’s permanent population. Level of educa-tional attainment: A = no formal education; B = elementary school; C = junior high school; D = seniorhigh school/technical school; E = college; F = university and above. Sources: Beijing Urban VillageSurvey; 2005 One-Percent Population Sample Survey in Beijing (Beijing Bureau, 2006).

ZHENG ET AL. 431


towns. The most popular provinces of origin are Henan, Hebei, Shandong, Anhui, andSichuan, which together account for 63 percent of migrants in the survey. Except for theirrank order, this top-five list is identical to the listing of top five origin provinces of migrantsaccording to the 2000 Census (Table 2, Fig. 7). Figure 7A, based on the 2000 Census, showsthe 14 largest interprovincial migration flows—i.e., streams of individuals moving to Beijingfrom another province over the past five years. Distance plays an important role, as provincesadjacent to Beijing, such as Hebei and Shandong, are among the major sources of migrantinflows. At the same time, non-adjacent provinces that are poor and have a long history ofsending out migrants, such as Sichuan and Anhui, also are major origin provinces.

Fig. 5. Income distribution of migrants. Source: Beijing Urban Village Survey.

Fig. 6. Comparison of migrants’ income with that of Beijing’s permanent labor force. Sources:Beijing Urban Village Survey; Beijing Bureau (2008).



In the survey, we asked migrants how long they intended to stay in Beijing. Among therespondents, 38.8 percent would like to remain permanently; 18.4 percent intend to stay foranother one to four years; 25.8 percent are prepared to leave at any time; and 17.1 percenthave “no idea” (see Fig. 8). These results suggest that the majority of migrant workers con-sider themselves as temporary, seeking primarily to earn money rather than to reside inBeijing permanently.

HOUSING IN URBAN VILLAGES

Urban villages in Chinese cities are well known for their poor physical environmentsand abundance of social problems. Our observations of the physical environment in the sur-veyed urban villages can be summarized as follows: houses are overcrowded; public stair-ways in built structures are very narrow; public facilities are inadequate and poorlymaintained; distances between houses are small and well below fire-control standards; andgarbage is scattered around (e.g., for three images from urban villages surveyed by our team,see Fig. 9). In short, the living conditions in urban villages are extremely poor, as reflected inthe migrants’ own perceptions. In our survey, 81 percent of migrants stated that they had per-sonally observed crimes in their neighborhoods during the six months prior to the survey.

Table 2. The Five Most Popular Origin Provinces of Migrants in Beijing, 2000–2008

2000 Census 2008 Urban Village Survey

Rank and province N Rank and province N

1. Hebei 373,810 1. Henan 1602. Henan 233,370 2. Hebei 1283. Anhui 150,390 3. Shandong 774. Shandong 145,870 4. Anhui 585. Sichuan 106,700 5. Sichuan 53

Sources: National Bureau, 2002 and Beijing Urban Village Survey.

Fig. 7. Origins of migrants to Beijing. Sources: A. National Bureau (2002). B. Beijing Urban Vil-lage Survey.

ZHENG ET AL. 433


High-density living is a hallmark of urban villages. Most of the housing units in the sur-veyed urban villages are one-room units, and the mean living space per dwelling room is

Fig. 8. Migrant workers’ intention to remain in Beijing.

Fig. 9. Urban villages in Beijing. A. Narrow alleys and balconies on upper floors (to maximizeliving space) are commonly seen in urban villages. B. A flyer on rooms available for rent is posted atthe entrance to a house. C. Urban villages of significant size may have their own supermarkets, laun-dry shops, vendors, and other stores. Some store owners are self-employed migrants.



13.2 m2. The mean per capita living space is 8.2 m2, less than one-third that (27.0 m2) inBeijing’s formal housing sector (houses built on state-owned land parcels) in 2007 (BeijingBureau, 2008). To control for location, for each surveyed urban village, we computed the

Table 3. Facilities in Rental Units in Beijing Urban Villages in 2008 (percent)

FacilityAvailability

FacilityAvailability

Yes No Yes No

Bathroom 4.2 95.8 Electricity 99.1 0.9Kitchen 8.1 91.9 Television 77.6 22.4Heating 14.0 86.0 Refrigerator 22.6 77.4Air conditioning 6.7 93.3

Source: Beijing Urban Village Survey.

Fig. 10. Unit sizes of urban village housing and adjacent formal rental housing. Sources: BeijingUrban Village Survey and Wo Ai Wo Jia (2009).

ZHENG ET AL. 435


average unit size of 10 adjacent formal rental housing units on the market. On average, theunit size ratio of urban village housing to formal rental housing is 1:6.32, confirming thelarge size gap between these two types of housing (Fig. 10). The two regressions in Figure 10show that although the unit size gradient with respect to distance to the city center for the for-mal housing sector has a slightly positive gradient, which is consistent with urban theory, thegradient for urban villages is flat. This means that migrants’ demand for unit size has no spa-tial variation—they consume the same amount of housing space everywhere.

Facilities in urban village units are inadequate and poorly maintained (Table 3). Morethan 90 percent of the surveyed units do not have bathrooms or kitchens. Dwellers of theseunits use public bathrooms and cook in public spaces. Despite the cold winters and summerheat in Beijing (daily minimums in January average 15.1°F and daily maximums in July87.4°F), 86.0 percent of the units we surveyed have no heating and 93.3 percent no air-conditioning. Almost all of the surveyed units do have electricity, however, and in fact televi-sion is the most popular electrical appliance (though most are old sets)—77.6 percent—in thesurveyed units; only 22.6 percent of the units have a refrigerator.

Fig. 11. Per-unit monthly rent in urban villages. Source: Beijing Urban Village Survey.



Despite the poor living conditions in urban villages, the low housing cost there hasattracted large numbers of migrants. According to our survey, the average monthly rent perurban village unit is 384.1 yuan, compared to approximately 2,000 yuan per one-bedroomunit in the formal housing sector. Because urban village units are mostly one-room unitswithout bathroom and kitchen, whereas units used for comparison in formal rental housingare primarily one-bedroom apartments with bathroom and kitchen, they are not exactly com-parable. Nonetheless, it is not surprising that the much higher per-unit rent in the formalhousing sector is too expensive for migrant workers.

Figure 11 and the regression associated with it examine the effect of distance to the citycenter on per-unit rent. Simple OLS regression indicates a significant negative rent gradientfor urban village units: an additional kilometer of distance from the city center reduces per-unit rent paid by 4.7 percent. This parsimonious model explains 34.6 percent of the spatialvariation in housing cost across urban villages.

We now further explore the roles of distance and unit size in determining rent. Figure 12shows that everywhere in Beijing the per unit rent in urban villages is significantly lowerthan that in adjacent formal housing units. However, the gap in rent per square meterbetween the two types of housing is small (Fig. 13). On average, rent per square meter inadjacent formal rental housing is only 1.23 times that of urban village housing, with severalurban villages showing rent per square meter equal to or even higher than that in adjacentformal rental housing. Thus, the two regressions associated with Figure 13 show that the (per

Fig. 12. Rent per dwelling unit for urban villages and adjacent formal rental units. Source:Beijing Urban Village Survey.

ZHENG ET AL. 437


square meter) rent gradients for urban village and formal housing sectors with respect to dis-tance to the city center are almost identical.

Comparing Figures 10–13 and their associated regressions, we can conclude that sizedifference is the key to explaining housing cost differentials between migrants and localresidents. In other words, migrants are willing to pay per-unit-space rent that is similar tothat in formal rental housing in similar locations, but their low income constrains theirchoices and reduces their space consumption. Urban villages that have small dwellingunits have therefore become the destinations of choice for migrants. This finding has ratherimportant policy implications. Knowing that migrants are willing to pay market or near-market per-unit-space rent, and that their size consumption is constrained by income, localgovernments can consider building affordable small-unit housing, in which facilities and

Fig. 13. Rent per square meter for urban villages and adjacent formal rental units. Source:Beijing Urban Village Survey.



amenities also are available; such housing units would provide an alternative to urban vil-lages for migrant housing.

MIGRANTS’ INCOME AND HOUSING

Modeling Wage and Housing Demand

It is well documented that housing consumption is a function of income. Standard hous-ing affordability studies have established 30 percent as the threshold of the income sharespent on housing (Bogdon and Can, 1997). In our survey described above, the income ofmigrant workers in urban villages is approximately one-half (51.2 percent) that of localworkers in Beijing, but the former’s housing consumption (in terms of housing cost) is lessthan 20 percent of their adjacent local residents. The share of housing cost in migrants’income is only 19.5 percent. In other words, not only can migrant workers in urban villagesafford their current rent, they can even afford a higher rent—i.e., more expensive and/orlarger dwellings. The question thus arises as to why more migrants do not move into the for-mal housing sector, which affords larger unit sizes as well as a more desirable living environ-ment. To better understand the underlying driving forces for migrants’ housing(neighborhood) choices, we explore the determinants of migrant workers’ wages and hous-ing demand. In so doing, we follow the standard labor economics literature to predict aworker’s expected earnings according to his/her human capital and demographics. TheMincerian wage equation, augmented with industry fixed effects, is:

log(WAGEi) = a0 + a1 · AGEi + a2 · AGEi2 + a3 · GENDERi + a4 · EDUi + a5 · BJYEAR + ß ×

industry_dummiesi + εi, (1)

where WAGE is migrant worker i’s monthly wage. We include the level and quadratic termsof the worker’s age (AGE, AGE2) to examine the nonlinear effect of age on wage. EDU refersto the worker’s educational attainment, measured by years of schooling. BJYEAR is the timelength (in years) the migrant worker has stayed in Beijing. Migrant workers accumulate workexperience and build social connections, which may bring in additional earnings. Therefore,we expect that the wage will rise with increasing duration of stay. For industries, we includenine categories—we selected wholesale and retail industry as the reference group anddefined the other eight industries as dummy variables. Table A2 in the Appendix providesthe descriptions and summary statistics for these variables.

Our housing demand equation is derived from household utility maximization under abudget constraint, which considers housing demand as a function of household income, thehousehold’s preference for housing, the cost of housing service, and the cost of other con-sumption goods. Housing demand is measured by total housing expenditure (Muth, 1960;Megbolugbe et al., 1991). For migrants in this study, housing expenditure is equivalent torent. The measure of income is dependent on housing tenure. For homeowners, householdincome is best represented by permanent income because housing is a durable good and thedecision to purchase a home is based on the household’s expected earnings when there is nocredit constraint (Polinsky and Ellwood, 1979). In our study, all migrant workers in urbanvillages are renters and are extremely credit-constrained, because they are not eligible formortgage loans from banks. Therefore, we use current income instead of permanent incomein the housing demand equation.

ZHENG ET AL. 439


Demographic characteristics are often included in the housing demand equation toreflect preference. In our study, we use age and household size as control variables for hous-ing preferences. When all observations are from the same city (or the same housingsubmarket), costs for housing service and other consumption goods are constant for theentire sample and are therefore not necessary for the model (Polinsky and Ellwood, 1979).For this reason, we do not include these housing service costs in the equation.

In addition to income, age, and household size, we also include the migrant worker’sexpectation for his/her future stay in Beijing. Our rationale is that if migrant workers con-sider themselves as temporary workers with the sole purpose of earning money before leav-ing in the future, they will save as much money as possible and will not be motivated toconsume, including housing consumption, in the city. Conversely, if migrants intend to stay,they may then be motivated to increase their housing consumption. Our survey includes aquestion on migrants’ intention to stay (see Fig. 8). In summary, our housing demand equa-tion (see Table A2) is:

log(RENTi) = γ0 + γ1 · HINCi + γ2 · AGEi + γ3 · HSIZEi + γ4 · STAY + δi, (2)

where RENT is migrant household i’s monthly rent expenditure, HINC is household monthlyincome, HSIZE is household size, and STAY is the migrant worker’s intention to stay inBeijing. The larger the value of STAY, the stronger the household’s intention to stay in Beijingin the future. We expect that this variable has a positive effect on housing consumption.

Empirical Results for the Wage Equation

Table 4 summarizes two regressions that estimate equation (1), which explain 22 to 25percent of the variation in migrant workers’ wage. Estimation (1) includes the demographicvariables, education measure, and industry dummies. EDU shows that the return rate to edu-cation is 3.1 percent—i.e., a one-year increase in schooling contributes to a 3 percent wageincrease. This is a much smaller rate than that of the overall labor force in Chinese cities,whose return to education is about 7 percent (Zheng et al., 2009). This gap in the economicreturn from education signals the persistence of labor market discrimination toward migrantworkers (Zhang and Meng, 2007).

Age has an inverted-U effect on wage, peaking at about 30 years. The initial positiveeffect of increasing age on wage reflects the accumulation of experience and skills withincreasing age. However, this effect is not linear and declines with increasing age, suggestingthat for migrant work that requires only low skills, physical ability is more often a determin-ing factor of wage level than experience. The wage of male migrant workers is 27 percenthigher than that of their female counterparts—a factor explaining why there are more menthan women living in urban villages (Table 1). Regarding industry dummies, we find thatmigrant workers in hotel/restaurant and service industries have the lowest earnings, whereasthose in culture industry command the highest wage. A detailed examination of our question-naire responses reveals that migrant workers in the service industry are overwhelminglyengaged in housekeeping jobs, which have low skill requirements.

In estimation (2), we include migrants’ duration of stay in Beijing (BJYEAR) in order toexamine their “learning curve” in cities. The learning effect is significant at the 1 percentlevel; a one-year increase in the duration of stay can increase the wage by 1.7 percent. Thisshows that migrant workers’ experience, working skills, and social connections in the city



can accumulate as their length of stay increases. This learning effect is about half the effectof formal education.

Empirical Results for the Housing Demand Equation

Table 5 summarizes the two regressions that estimate Equation (2). Estimation (1)includes household income (in logarithmic scale), age, and household size. Generally, housingdemand increases with both household income and household size. Adding one more memberto a household increases housing demand by about 17.4 percent. AGE does not have a signifi-cant effect on housing demand.4 The income elasticity of housing demand—the coefficient ofLog(HINC)—is 0.129, indicating that a 10 percent increase in household income will result ina 1.29 percent increase in housing demand. This elasticity is much lower than the income elas-ticity of 0.499 for the renters in Chinese cities as a whole (Zheng, 2007). The low incomeelasticity of housing demand for migrant workers reflects their weak motivation to spend(especially on housing) in cities, supporting our observation made earlier.

In Estimation (2), we add STAY (intention to stay in Beijing) in order to examine whethermigrant workers who plan to stay in Beijing for many years are more likely to rent larger or

Table 4. Regression Results for the Wage Equationa

Variable

Log(WAGE)

(1) (2)

Coeff. (t-stat.) Coeff. (t-stat.)

AGE 0.063 (5.4)*** 0.046 (3.9)***AGE2 -0.001 (6.2)*** -0.001 (4.9)***GENDER 0.266 (6.2)*** 0.250 (5.8)***EDU 0.031 (4.8)*** 0.031 (4.9)***BJYEAR 0.017 (4.5)***Industry dummies:MANUFACTURING -0.109 (1.5) -0.098 (1.4)CONSTRUCTION 0.089 (1.5) 0.063 (1.0)TRANSPORT 0.012 (0.2) 0.016 (0.2)HOTEL_RESTRT -0.132 (2.1)** -0.135 (2.2)**REALESTATE 0.159 (0.7) 0.164 (0.8)SERVICE -0.240 (4.2)*** -0.236 (4.1)***CULTURE 0.242 (1.7)* 0.184 (1.3)OTHER_IND -0.101 (0.9) -0.081 (0.7)constant 6.091 (28.0)*** 6.311 (28.6)***Observations 752 748R2 0.224 0.246

aThe number of observations is not equal to the survey sample size due to missing values for WAGE and BJYEAR;***, **, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively.

4We also tried AGE2, but found that it also was not significant.

ZHENG ET AL. 441


better homes. This variable has a positive and significant coefficient (at the 1 percent level),further confirming the relationship between intention to stay and the consumption of hous-ing.

Figure 14 summaries how migrant workers’ household income is used. As noted previ-ously, housing expenditure accounts for only 19.5 percent of migrant’s household income.The ratio of total expenditure to income is only 53 percent for migrants, compared to 72.5percent for urban residents (National Bureau, 2008). As discussed earlier, from the housingaffordability perspective these migrants can not only afford their current rental units in urbanvillages but actually can afford even better housing. Nonetheless, migrant workers do notchoose to improve their housing consumption. Instead, they save almost half of their earn-ings, which are customarily sent back as remittances to their rural homes (Fig. 14). Theabove therefore suggests that migrant workers live in crowded and cheap urban villagehouses not because they cannot afford better housing accommodation, but because they arenot willing to spend their hard-earned income in cities.

Table 5. Regression Results for the Housing Demand Equationa

Variable

Log(RENT)

(1) (2)

Coeff. (t-stat.) Coeff. (t-stat.)

Log(FINC) 0.129 (3.7)*** 0.086 (2.2)**AGE -0.001 (0.7) -0.002 (0.9)HSIZE 0.174 (10.8)*** 0.180 (10.4)***STAY 0.041 (2.6)***Constant 4.562 (16.0)*** 4.784 (15.2)***Observations 697 586R2 0.178 0.186

aThe number of observations is not equal to the survey sample size due to missing values for WAGE and BJYEAR.***, **, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively.

Fig. 14. Usage of migrant workers’ household income.



Migrant workers’ reluctance to improve their housing consumption in cities clearlyimplies that to them the city is only a place for work and not a permanent residence. A majorreason for this observation is the institutional barriers migrants face in cities. Despite relax-ations of mobility constraints that since the 1980s have made possible increases in rural labormigration, it is still extremely difficult for rural Chinese to obtain urban hukou (Chan, 2009).The bundle of urban social services (such as public housing, health insurance, unemploymentinsurance, pension fund, housing provident fund, etc.) is closely tied to the hukou system.The vast majority of migrant workers in cities are seen and considered as temporary workersand thus do not have access to these social services and many urban amenities (Fan, 2001,2002). Rather than seeing cities as their homes, migrants circulate between urban work andthe countryside, and it is the home village and community that constitutes their long-termsocial and economic security (Fan, 2008; Fan and Wang, 2008). It is not surprising, therefore,that migrant workers are cautious about plans to stay and consume in the city (Zhu, 2007;Zhu and Chen, 2009).

SUMMARY AND CONCLUSION

Urban villages epitomize not only the landscapes but also the complex processes ofrural-urban migration in China. Many rural villages situated on the fringes of cities havebeen encroached upon by urban expansion and are inhabited by large numbers of migrantworkers. Using a recent survey of 50 such urban villages in Beijing, we have documented inthis paper the living and work conditions of migrant workers and analyzed their housing con-sumption behavior.

Our survey reveals demographic and social characteristics of migrant workers that aresimilar to those described in existing studies. The majority of migrants living in urban vil-lages are male and young. Most have low levels of educational attainment and hold low-skilljobs in the tertiary sector. Their average wage is roughly half of that of the local labor force.In addition to demographic and social characteristics, our study has yielded fresh anddetailed information about the living conditions of migrant workers. Their housing condi-tions are poor: rental units in urban villages are overcrowded, facilities are inadequate andpoorly maintained, and social problems are rampant. Migrant workers’ housing consumptionbehavior is marked by, first, a willingness to pay per-unit-space rent at or near market levelsas reflected in adjacent formal rental units, and second, small space consumption and accord-ingly low overall expenditure on rent. And, unlike formal rental units that increase in sizewith increasing distance from the city center, dwelling units of migrants in urban villages areof similar size across the city. In other words, migrants are attracted to urban villages notbecause the dwelling units there have lower-than-market rent but because they are small.

An obvious reason for migrants’ small size consumption is their low income. To ascer-tain that relationship, we have investigated the determinants of migrant workers’ wage andhousing demand by estimating the Mincerian wage equation and a housing demand equation.The wage equation reveals a small return on schooling for migrant workers, signaling persis-tent discrimination in the urban labor market. We have found a significant learning curve forrural migrants, suggesting that cities with agglomeration economies will provide learningbenefits for migrants who have remained there for a long period of time. This learning bene-fit is likely to sustain continued rural-urban migration in the future.

Our housing demand equation shows that the income elasticity of housing demand formigrants is much lower than that of local residents in cities. On average, migrants spend only19 percent of their income on housing and save about half of their income. We have also

ZHENG ET AL. 443


found that only those who plan to stay in Beijing in the future are willing to rent larger or bet-ter homes.

The above findings suggest that migrant workers choose urban village housing that issmall, overcrowded, and poorly served not because they cannot afford better housing butbecause they are not willing to consume in the city. Due to barriers to urban hukou and asso-ciated benefits, labor market discrimination, and social segregation and inferiority, mostmigrant workers consider the city as merely a place to work but not to live. As a result, theyare motivated to circulate between sites of migrant work and their home villages, and to savetheir income or expend it on rural households in their home villages rather than on consump-tion in the city.

From a theoretical perspective, our findings highlight the stickiness of rural-urbanmigration. Rather than a one-way flow from the countryside to the city, rural-urban migrationin China is marked by long-term circulatory movements, barriers to permanent urban settle-ment, and migrants’ difficulties and reluctance to “become urban” other than working for awage in the city. From a policy perspective, our study shows that migrants are willing to paymarket rent as long as the dwelling unit is small and thus the total housing expenditure also issmall. Local governments can consider building affordable small-unit housing, where facili-ties and amenities are available, as an alternative to urban villages for migrant housing. Poli-cies that improve the job and social security of migrant workers would increase migrantworkers’ willingness to spend more on housing, which would in turn improve their livingconditions.

The current global financial crisis has resulted in millions of migrant workers losingtheir jobs. Many have returned to their original villages to farm, but others insist on lookingfor migrant work. These recent events reinforce the notion that for migrant workers the cityis merely a place to work but not to live. When the job opportunities are gone, migrants havefew options other than to leave the city. In order for these workers to stay and spend (forexample, on housing) in cities, policies that enable them to accumulate human capital, takehigh-skill jobs, and increase social integration are necessary.

REFERENCES

Beijing Bureau of Statistics, Beijing Tongji Nianjian 2006 (Beijing Statistical Yearbook 2006).Beijing, China: China Statistics Press, 2006.

Beijing Bureau of Statistics, Beijing Tongji Nianjian 2008 (Beijing Statistical Yearbook 2008).Beijing, China: China Statistics Press, 2008.

Bogdon, Amy S. and Ayse Can, “Indicators of Local Housing Affordability: Comparative and SpatialApproaches,” Real Estate Economics, 25, 1:43–80, 1997.

Chan, Kam Wing, “The Chinese Hukou System at 50,” Eurasian Geography and Economics, 50,2:197–221, 2009.

Fan, C. Cindy, “Migration and Labor Market Returns in Urban China: Results from a Recent Survey inGuangzhou,” Environment and Planning A, 33, 3:479–508, 2001.

Fan, C. Cindy, “The Elite, the Natives, and the Outsiders: Migration and Labor Market Segmentationin Urban China,” Annals of the Association of American Geographers, 92, 1:103–124, 2002.

Fan, C. Cindy, China on the Move: Migration, the State, and the Household. London, UK and NewYork, NY: Routledge, 2008.

Fan, C. Cindy and Wenfei Winnie Wang, “The Household as Security: Strategies of Rural–UrbanMigrants in China,” in Russell Smyth and Ingrid Nielsen, eds., Migration and Social Protection inChina. Hackensack, NJ: World Scientific, 2008, 205–243.



Guangming ribao, “Zhou Yixing: Chengzhenhua bushi yuekua yuehao (Zhou Yixing: Rapid Urban-ization Is Not Necessarily Good),” Guangming ribao, March 27, 2006.

Hsing, You-tien, The Great Urban Transformation: Politics and Property in China. Oxford, UK:Oxford University Press, 2009.

Jacobs, Jane, The Death and Life of American Cities. New York, NY: Random House, 1961.Lin, George C. S., “Peri-urbanism in Globalizing China: A study of New Urbanism in Dongguan,”

Eurasian Geography and Economics, 47, 1:28–53, 2006.McGee, T. G., George C. S. Lin, Andrew M. Marton, Mark Y. L. Wang, and Jiaping Wu, China’s

Urban Space: Development Under Market Socialism. London, UK and New York, NY: Routledge,2007.

Megbolugbe, Isaac F., Allen P. Marks, and Mary B. Schwartz, “The Economic Theory of HousingDemand: A Critical Review,” Journal of Real Estate Research, 6, 3:381–393, 1991.

Mobrand, Erik, “Politics of Cityward Migration: An Overview of China in Comparative Perspective,”Habitat International, 30, 261–274, 2006.

Muth, Richard F., “The Demand for Nonfarm Housing,” in Arnold C. Harberger, ed., The Demand forDurable Goods. Chicago, IL: University of Chicago Press, 29–96, 1960.

National Bureau of Statistics, Beijing 2000 Nian Renkou Pucha Ziliao (Beijing 2000 Population Cen-sus Materials). Beijing, China: China Statistics Press, 2002.

National Bureau of Statistics, Zhongguo Tongji Nianjian 2008 (China Statistical Yearbook 2008).Beijing, China: China Statistics Press, 2008.

Polinsky, A. Mitchell and David T. Ellwood, “An Empirical Reconciliation of Micro and GroupedEstimates of the Demand for Housing,” The Review of Economics and Statistics, 61, 2: 199–205,1979.

Song, Yan, Yves Zenou, and Chengri Ding, “Let’s Not Throw the Baby out with the Bath Water: TheRole of Urban Villages in Housing Rural Migrants in China,” Urban Studies, 45, 2:313–330, 2008.

Tang, Wing-Shing and Him Chung, “Rural–Urban Transition in China: Illegal Land Use and Con-struction,” Asia Pacific Viewpoint, 43, 1:43–62, 2002.

Tian, Li, “The Chengzhongcun Land Market in China: Boon or Bane?—a Perspective on PropertyRights,” International Journal of Urban and Regional Research, 32, 2:282–304, 2008.

“Wo Ai Wo Jia (I Love My Home),” 2009 [http://www.5i5j.com], accessed April 2009.Wu, Fulong, “Beyond Gradualism: China’s Urban Revolution and Emerging Cities,” in Fulong Wu,

ed., China’s Emerging Cities: The Making of New Urbanism. London, UK and New York, NY:Routledge, 2007, 3–25.

Zhang, Dandan and Xin Meng, “Assimilation or Disassimilation? The Labour Market Performanceof Rural Migrants in Chinese Cities,” in Proceedings of the Sixth Conference on the Chinese Econ-omy. Clermont-Ferrand, France: CERDI-IDREC, 2007, 18–19.

Zhang, Kevin Honglin and Shunfeng Song, “Rural–Urban Migration and Urbanization in China:Evidence from Time-Series and Cross-Section Analyses,” China Economic Review, 14, 4:386–400, 2003.

Zhang, Li, Simon X. B. Zhao, and J. P. Tian, “Self-Help in Housing and Chengzhongcun in China’sUrbanization,” International Journal of Urban and Regional Research, 27, 4:912–937, 2003.

Zhang, Yan, “Planners Urged to Tackle Problem of Villages Swallowed up by Urban Sprawl,”China.Org.Cn, March 16, 2002 [http://www.china.org.cn/english/2002/Mar/28926.htm], accessedApril 2009.

Zheng, Siqi, Zhufang Xuqiu De Weiguan Jingji Fenxi (A Microeconomic Analysis of Housing Demandin China). Beijing, China: China Architecture and Building Press, 2007.

Zheng, Siqi, Yuming Fu, and Hongyu Liu, “Demand for Urban Quality of Living in China: Evolutionin Compensating Land-Rent and Wage-Rate Differentials,” Journal of Real Estate Finance andEconomics, 38, 3:194–213, 2009.

Zhou, Yixing and John R. Logan, “Growth on the Edge: The New Chinese Metropolis,” in John R.Logan, ed., Urban China in Transition. Malden, MA: Blackwell, 2008, 140–160.

ZHENG ET AL. 445


Zhou, Yixing and Laurence J. C. Ma, “China’s Urban Population Statistics: A Critical Evaluation,”Eurasian Geography and Economics, 46, 4:272–289, 2005.

Zhu, Yu, “China’s Floating Population and Their Settlement Intention in the Cities: Beyond the HukouReform,” Habitat International, 31, 1:65–76, 2007.

Zhu, Yu and Wenzhe Chen, “The Settlement Intention of China’s Floating Population in the Cities:Recent Changes and Multifaceted Individual-Level Determinants,” Population, Space and Place,15, 2009 (forthcoming).

APPENDIX

Table A1. The 50 Surveyed Urban Villages

Village Natives Migrants Migrant/native ratio Village Natives Migrants Migrant/

native ratio

Bajia 1,700 45,000 26.5 Cuigezhuang 1,683 35,000 20.8Magezhuang 3,000 50,000 16.7 Leijiaqiao 1,087 10,000 9.2Beishuang qiao 405 1,310 3.2 Jiugong 323 3,500 10.8Zhongtan 1,500 20,000 13.3 Shoubaozhuang 800 7,800 9.8Dongxiaokou 1,730 15,000 8.7 Majiapu 2,788 780 0.3Maliandian 1,650 14,000 8.5 Dahongmen 5,500 6,000 1.1Yamenkou West 2,528 3,397 1.3 Yushuzhuang 2,300 1,000 0.4Beimafang 2,040 3,000 1.5 Dajing 15,000 7,000 0.5Caogezhuang 1,100 3,500 3.2 Sanluju 400 1,000 2.5Mafang 4,500 50,000 11.1 Beihuqu 320 1,200 3.8Dongsanqi 1,860 27,000 14.5 Changpo 1,125 900 0.8Huilongguan 3,300 28,000 8.5 Xiaowuji 3,800 20,000 5.3Wangsiying 1,037 12,000 11.6 Beihuayuan 2,500 2,000 0.8Kuifang 4,500 10,000 2.2 Tuofangying 6,100 8,000 1.3Liulangzhuang 3,000 20,000 6.7 Xizhihe 3,000 3,000 1.0Shucun 6,600 50,000 7.6 Gaojing 3,480 16,138 4.6Niuwangmiao 1,000 3,000 3.0 Sum 91,656 478,525 5.2

Villages for which migrant and native numbers are not available

Fenzhongsi TiancunGaomiao WanjuzhuangHeizhuanghu WudaokouHuangtingzi XinfadiKandanqiao XinshengLiujiayao YaojiayuanNanyuan YujiafenShiliheng YuquanTishi West



Table A2. Variable Definitions and Summary Statistics

Variable Description of migrant worker Mean Standarddeviation

WAGE Monthly wage (yuan) 1984.4 1079.5HINC Monthly household income (yuan) 2715.2 1493.5AGE Age (years) 33.9 9.9GENDER Binary, gender, 1 = male 0.669 0.471EDU Length of schooling (years) 8.7 3.2HSIZE Household size (people) 2.3 1.2BJYEAR Length of stay in Beijing (years) 7.1 5.5MANUFACTURING Binary, 1 = employed in manufacturing. 0.090 0.286CONSTRUCTION Binary, 1 = employed in construction 0.148 0.355

TRANSPORT Binary, 1 = employed in transport, storage, and postal 0.075 0.264

SALE Binary, 1 = employed in wholesale and retail trade 0.339 0.474HOTEL_RESTRT Binary, 1 = employed in hotel and restaurants 0.127 0.333REALESTATE Binary, 1 = employed in real estate 0.008 0.089

SERVICE Binary, 1 = employed in household and other services 0.157 0.364

CULTURE Binary, 1 = employed in culture, sports, and entertainment 0.020 0.140

OTHER_IND Binary, 1 = employed in other than above eight industries 0.032 0.175

STAY

Intention to stay in Beijing: 1 = will leave any time; 2 = will remain in Beijing for another one or two years; 3 = will remain in Beijing for another three or four years; 4 = will remain in Beijing permanently; NULL = no idea.

2.7 1.3

CONSUMPTION Household consumption expenditure (yuan) 1405.1 987.3RENT Monthly rent (including expenditure on facilities) 444.6 373.0

Documents

Urban Villages in China: A 2008 Survey of Migrant … ET AL. 427 better understanding of the housing demand a nd housing choice behaviors of migrants is cru-cial for developing effective